^{page 247 note 1} The “fire-pit” legend first appears in the Pṛithvīrāj-rāisā of Cand (Chand), which claims to be contemporary with the hero of the poem, but may be considerably later in date. Pṛithvīrāja was killed in 1193 A.D.

^{page 249 note 1} “Epigraphic Notes and Questions,” p. 18Google Scholar of reprint from J. Bo. Br. R.A.S., vol. xx.Google Scholar

^{page 250 note 1} Bāṇa, , Harṣa-carita, transl. Cowell, & Thomas, , p. 116.Google Scholar See Hoernle, , ante, p. 137.Google Scholar

^{page 251 note 1} The three grants are (1) the Waṇi grant dated, according to Dr. Fleet, in “Śaka 730 for 728 (A.D. 806–7, the Vyaya saṁvatsara” (Ind. Ant., xi, 156, 160Google Scholar); (2) the Rādhanpur grant, dated 730 Śaka, 808 A.D. (Kielhorn, , Ep. Ind., vi, 240Google Scholar); and (3) the Barodā grant of Kakkarāja II, dated Śaka 734, 812–13 A.D. (Fleet, , Ind. Ant., xii, 164Google Scholar). The text of the Waṇi grant, repeated in the Rādhanpur document, is as follows:—“Having with his armies, which no other army could withstand, quickly caused Vatsarāja, intoxicated with the goddess of the fortune of (the country of) Gauḍa, that he had acquired with ease, to enter upon the path of misfortune in the centre of (the deserts of) Maru, he took away from him not only the two (regal) umbrellas of Gauḍa, that were as radiantly white as the rays of the autumn moon, but also, at the same moment, his fame that had reached to the extremities of the regions.” Kakkarāja's grant explains that the entrance to Mālava was barred against the lord of the Gurjaras, who had been “evilly inflamed by conquering the lord of Gauḍa and the lord of The Rāṣṭrakūṭas being the enemies of the Gurjaras, the testimony of the Rāṣṭrakūṭa grants to the success gained by Vatsarāja in his war against Gauḍa may be accepted without reserve. The mention of both Gauḍa and in Kakkarāja's grant explains the reference to two state umbrellas in the Waṇi and Rādhanpur documents.

^{page 253 note 1} Bom. Gaz., 1896, vol. i, pt, ii, p. 197Google Scholar n.

^{page 254 note 1} Ep. Ind., vi, 195.Google Scholar

^{page 255 note 1} It is unnecessary to discuss the early tentative interpretations of the passage, according to which Vatsarāja was supposed to mean “King of the Vatsas”, and so forth. It seems to me quite clear that the epithet Kṛṣṇaṇripaje was intended to define the application of the indeterminate title Śrī Vallabha. Dr. Fleet long maintained that Govinda II never actually reigned, but the correctness of the contrary opinion has been demonstrated successfully by Mr. D. R. Bhandarkar in his paper on the Cambay plates of Govinda IV (Ep. Ind., vii, 28Google Scholar).

^{page 256 note 1} Unpublished inscription from Buckalā in Jodhpur State, No. 2 of my List. (Now ed. by Bhandarkar, D. R., Ep. Ind., ix, 198.Google Scholar)

^{page 256 note 2} Annual Report, Archœol. Survey, 1903–1904, p. 279.Google Scholar

^{page 258 note 1} “Ānarta is known from the Mahābhārata and the Purāṇas. It corresponds to northern Kāṭhiāvād. Its capital was Kuśasthalī, the modern Dvārkā” (Indrajī, Bhagwān Lal and Bühler, , Ind. Ant., vii, 259Google Scholar). Benfey refers to Mbh., 2, 614.Google Scholar

According to Cunningham, “Matsya then included the whole of the present Alwar territory, with portions of Jaypur and Bharatpur. Bairāt and Māchāri were both in Matsya-desa; while Kāman, Mathurā, and Bayāna were all in Surasena” (Reports, xx, 2;Google ScholarAnc. Geogr., pp. 340–5Google Scholar).

“De l'ensemble des témoignages, il apparaît qu'aux temps anciens, les Hindous désignaient sous le nom de Kirâtas toutes les populations de famille tibéto-birmane qui s'échelonnaient entre les hauts plateaux de l'Himalaya, les bouches du Gange et le littoral voisin. Refoulés ou absorbés par la poussée hindoue, les Kirâtas n'ont subsisté que dans les montagnes, à l'Est du Népal” (Lévi, Sylvain, Le Népal, vol. ii, p. 77Google Scholar). But the term is also used in a less definite sense. The names Kirāta and Pulinda are employed interchangeably in the Daśakumāracarita (Bomb, . ed., p. 25).Google Scholar Dr. Collins holds, and apparently with reason, that the name Kirāta “can only be regarded as one of the general terms for non-Aryan aborigines” (Geographical Data of the Raghuvaṁśa and Daśakumāracarita, Leipzig, 1907, pp. 8, 18).Google Scholar

Vatsa is said to have been in the south-eastern division of the middle country (Bṛhat Saṁhitā transl. in Ind. Ant., xxii, 193Google Scholar). For discussion of the position of Kauśāmbī, see JRAS., 1898, p. 503;Google Scholar 1900, p. 1; 1903, p. 583; 1904, pp. 249, 544.

^{page 259 note 1} The Bhāgalpur copper-plate records a grant made by Nārāyaṇa-pāla, great-grandson of Vākpāla, brother of Dharmapāla. Verse 3 recites that Dharmapāla gave the sovereignty over Mahodaya or Kanauj, which he had acquired by the defeat of Indrarāja (= Indrāyudha) and others, to the suppliant Cakrāyudha (Ind. Ant., xv, 304;Google Scholar xx, 188).

The Khālimpur copper-plate, dated in the year 32 of the reign of Dharmapāla, recites (v. 12) that he had “installed the illustrious king of Kanyakubja, who readily was accepted by the Bhoja, Matsya, Madra, Kuru, Yadu, Yavana, Avanti, Gandhāra, and Kīra kings, bowing down respectfully with their diadems trembling, and for whom his own golden coronation jar was lifted up by the delighted elders of Pañcāla” (Ep. Ind., iv, 252Google Scholar). Although this contemporary record does not name the king of Kanauj, it is obvious that the event described is the installation of Cakrāyudha in succession to Indrarāja (= Indrāyudha), as recorded in the later Bhāgalpur grant.

The position of the Bhoja kingdom has been discussed recently with much learning by Dr. Mark Collins, who has shown good reason for believing that the Vidarbha of Kālidāsa's Raghuvaṁśa and Dandin's Daśakumāracarita should be identified with the kingdom of the Vākāṭaka dynasty, the members of which were feudatories of the later kings of the imperial Gupta line. The rulers of this kingdom were known as Bhojas, and the political centre of their dominion was Bhojakaṭarājyam, corresponding roughly with the Ilicpur District in Berār, to the north of the Tapti (Tāpī) River. The authority of the Vākāṭakas in the sixth century had extended certainly as far north as the native state of Jaso in the Bundelkhaṇḍ group of Central Indian States (circâ lat. 24° 24′ long. 80° 30′), on the northern side of the Narmadā, and it is quite possible that Bhojas from Vidarbha may have attended the installation of Cakrāyudha. When that event took place the Vākāṭaka dynasty had long become extinct, but the tribal name Bhoja survived, and the Bhojas who came into touch with Dharmapala most likely lived near the Narmadā.

The Matsya country, as already noted, is supposed to be equivalent to Eastern Rājputāna. The Madra nation certainly occupied the central parts of the Panjāb. The Kurus may be associated with the Kurukshetra of Thānēsar, and the Yadus, or Yādavas, may be the people of the Mathurā territory. The name Yavana, as is well known, was used in an indefinite way to designate various races of foreign origin in the north-west of India, and it is impossible to identify precisely the meaning attached to the term by Dharmapāla's historiographer. Avanti, of course, means Eastern Mālwā, or the Ujjain territory. Gandhāra, presumably, has its usual meaning, and designates the Peshāwar District, with certain adjoining regions. The Kīras occupied territory in the Kāngṛā Valley, where their name is preserved at Kīragrāma, the village in which the famous temple of Vaidyanāth or Baijnāth stands, or stood until the earthquake of 1905. One of the inscriptions there shows that in the eleventh century (926 Śaka, corrected date) Kīragrāma was a fief of the kingdom of Trigarta or Jālandhara. In literature the Kīras are often mentioned in connexion with the Hūnas, and also with the Kulūtas, or inhabitants of Kulu (Ep. Ind., i, pp. 97 ff.Google Scholar; JRAS., 1900, p. 540;Google ScholarCunningham, , Reports, v, 178;Google ScholarArchœol. Surv. Annual Rep., 1902–1903, p. 268Google Scholar).

The defeat of Nāgabhaṭa by the Rāṣṭrakūṭa king, Govinda III, is mentioned in an unpublished inscription of Amoghavarṣa belonging to Professor R. G. Bhandarkar (Bhandarkar, , “The Gurjaras,” p. 4Google Scholar of reprint).

^{page 261 note 1} Schiefner could not explain this phrase. The pillar has not been discovered. The mention of Delhi is an anachronism, as the city was not founded until 993–4 A.D. (E. Hist. India, 2nd ed., p. 355Google Scholar), but the historian intends only to mark the western limit of Dharmapāla's rule by naming a well-known place.

^{page 261 note 2} Khri srong lde btsan was the patron of Guru Padma-sambhava, a native of Udyāna, trained at Nālanda, who came to Tibet in 747 A. D. and introduced Lamaism (Waddell, , The Buddhism of Tibet, 1895, p. 24Google Scholar). The dates given in the text for the birth and reign of Khri srong Ide btsan rest on the authority of Sarat Candra Dās (“Contributions on Tibet,” JASB., 1881, vol. iv, pt. i, p. 224).Google Scholar

^{page 263 note 1} Ind. Ant., xii, 189;Google Scholar No. 77 of Kielhorn's, “Southern List”Google Scholar (Ep. Ind., vii,Google Scholar App.).

^{page 263 note 2} Cunningham, , Coins of Med. India, p. 49,Google Scholar pl. vi, 20, 21; Catal. Coins, I.M., vol. i, pp. 233, 241,Google Scholar pl. xxv, 18.

^{page 264 note 1} Bom. Gaz., 1896, vol. i, pt. i, pp. 468, 526.Google Scholar

^{page 265 note 1} Elliot, , i, 4.Google Scholar

^{page 266 note 1} Kielhorn, , Ep. Ind., ix, pp. 1–4.Google Scholar While revising the draft of this essay I have heard with the deepest regret the news of the sudden death of Professor Kielhorn, whose labours have made it possible for me to give a detailed history of the Gurjara-Pratihāra empire.

^{page 267 note 1} In Kaśmīr the term viṣaya=pargana=the later term rāṣṭra (Stein, , transl. Rajatar., vol. ii, p. 437Google Scholar). This meaning of viṣaya certainly suits many inscriptions. The pargana is a wonderfully stable unit of high antiquity. Maṇḍala means a larger area, more or less equivalent to the modern district, or perhaps division. Bhukti denotes an extensive province or kingdom, as Jejākabhukti = Bundelkhaṇḍ; Tīrabhukti = Tirhūt. The reader will observe that the ancient names of three viṣayas, viz., Vālayikā, Kuṇḍadhāni, and Śrāvastī, included in the Śrāvastī province, are recorded. Probably they are still traceable. (Now see, for a different view, Vogel, , ante, 1908, p. 971.Google Scholar)

^{page 269 note 1} In Rājputāna, in Malcolm's time, every male Rājput infant received, as the first present from his father, an embossed figure of a horse and the sun, made of either gold or silver, which he continued to wear suspended from his neck, worshipping it daily as his personal deity (Malcolm, , Central India, 1832, 3rd ed., vol. ii, p. 144Google Scholar).

^{page 269 note 2} JRAS., N.S., vol. xiv, p. 319,Google Scholar quoted by Fleet in Kanarese Dynasties (Bom. Gaz., vol. i, pt. ii, p. 380Google Scholar), and by Jackson, ibid., pt. i, pp. 466, 469. The passage runs, as given by Fleet:—His son Narasiṁha … plucked the goddess of victory from the arms of Ghūrjararāja (sic), defeated a king named Mahīpāla, and bathed his horses at the junction of the Ganges.” This seems to mean, as interpreted by Mr. Jackson, that Mahīpāla is described as “the king of the Gurjaras ”.

^{page 271 note 1} Elliot, , Hist. of India, vol. i, p. 25;Google ScholarBom. Gaz., 1896 vol. i, pt. i, pp. 469, 506, 525.Google Scholar

^{page 272 note 1} Ep. Ind., i, 122.Google Scholar The fragments of lines 7 and 8 describe Harṣa as a conqueror, and what is left of line 10 consists of the words punaryena Śrī Kṣitipāladēvaṇṛipatiḥ siṁhāsane sthā[pitaḥ]. The difficulty consists in determining the antecedent of the relative pronoun yena, “by whom.”

^{page 272 note 2} Probably the Kāngṛā Valley.

^{page 274 note 1} Ind. Ant., xix, 233;Google ScholarJackson, , in Bom. Gaz., 1896, vol. i, pt. i, pp. 451, 469.Google Scholar By a clerical error the date of Mūlarāja is given as 740 in the note on p. 451.

^{page 274 note 2} Khajurāho inscription No. 2, vv. 44, 45:—“ … who rules the earth … even as far as that mountain called Gopa (Gopādri)” (Ep. Ind., i, 134Google Scholar). For limits of dominions see ibid., p. 124. The Jumna was the frontier between them and the Kanauj kingdom of Pañcāla.

^{page 274 note 3} The relation of feudatory is an inference from the Khajurāho inscription. The only inscription of Vajradāman is dated in 977 A.D. (No. 47 of Kielhorn's “Northern List”; JASB., vol. xxxi, p. 393).Google Scholar The conquest by Vajradāman is recited in a much later Gwālior inscription dated 1150 V.E. = approximately 1093 A.D. (Ind. Ant., xv, 35, 41;Google ScholarKielhorn's, “Northern List”,Google Scholar No. 73).

^{page 276 note 1} Raverty, , Notes on Afghānistān, p. 320.Google Scholar That author shows good reason for holding that the battle took place in or near the Kurram Valley.

^{page 279 note 1} The narrative in the text is based on a combination of the testimony of the inscriptions with that of the Muhammadan historians. Al-'Utbī, author of the was Secretary to the Sultan Maḥmūd, and is, therefore, entitled to credit as a contemporary authority. With one exception, his narrative does not deal with events subsequent to 1020 A.D. (410 H.). He records briefly the hardly won victory gained by Maḥmūd over Ānandapāla, king of the Panjāb, late in 1008 A.D., and states that the allies were led by Brāhmaṇapāla, son of Ānandapāla (Elliot, , ii, 33Google Scholar). A fuller account is given by the late compiler, Firishta, who presumably based his narrative on earlier authorities (Elliot, , ii, 446Google Scholar).

Al-'Utbī's history of the Kanauj expedition is the best that we possess (Elliot, , ii, 41, 42, 44–6, 457Google Scholar). The Sultan, having crossed the Jumna (Jūn) on the 20th Bajab, 409 H. = December 2, 1018 A.D., then captured various strongholds, sacked Mathurā, and advanced on Kanauj. “He left the greater part of his army behind, and took only a small body of troops with him against Rāi Jaipāl [v.l. Rājāpāl, Rajaipāl, etc., slight corruptions of Rājyapāla], who had also but a few men with him, and was preparing to flee for safety to some of his dependent vassals.”

When Elliot, Reinaud, and other authors commented on Al-'Utbī's narrative the Parihār and Candēl inscriptions were not known, and “Rāi Jaipāl” was an insoluble puzzle. He was confounded with Jaipāl, the king of the Panjāb, who had opposed Sabuktigīn, and was also mixed up with the princes of the Ṣāhiya dynasty of Ohind (Kābul). Now that it is certain that Rājyapāla was the Rājā of Kanauj in 1018 A.D., it is easy to make the slight correction required in Al-'Utbī's text, and the whole story becomes fully intelligible.

the author of the seems to be in error in representing the sack of Mathurā as subsequent to that of Kanauj (Elliot, , ii, 460Google Scholar). The testimony of the contemporary Al-'Utbī is to be preferred. The explanations that Rājyapāla's death was a punishment inflicted by his allies for his failure to resist the invader and that Maḥmūd returned to India with the special purpose of punishing the Hindu allies in his turn are due to Niຓām-ud-dīn. By the accidental omission of a stroke the name of Gaṇḍa, is transformed into Nandā, in the Persian texts. The true form of the name is known from the Mau Candēl inscription (Ep. Ind., i, 295–307Google Scholar). The Dūbkuṇḍ inscription near Gwālior (Ep. Ind., ii, 235Google Scholar) records the slaying of Rājyapāla, king of Kanauj, by Arjuna, the Kacchwāha chief of Gwālior, under the command of Vidhyādhara Candēl, who must have been then Crown Prince, as his father Gaṇḍa was still alive when attacked later by Maḥmūd. Another inscription found at Mahobā refers to the same event by describing Vidhyādhara as “a master of warfare, who had caused the destruction of the king of Kanyākubja” (Ep. Ind., i, 219Google Scholar).

Inscription No. 22 in the list in this paper, dated January 26, 1027 A.D., establishes the fact that in the Parihār dynasty of Kanauj Trilocanapāla was the successor of Rājyapāla, the successor of Vijayapāla of the same dynasty, who was reigning in 960 A.D. The name of Trilocanapāla also occurs in where it has not been recognized hitherto. Describing the invasion of 410 H. = 1019–20 A.D., the historian says that when Maḥmūd “reached the banks of the Jumna, Tarū Jaibāl [reading of Elliot's MS.], who had so often fled before his troops, and who had now come to assist Nandā, encamped in face of the Sultan”, etc. It is now obvious that the reading “Tarū Jaibāl” is merely a slight error for Trilocanapāla, the dropping of the upright stroke of the l (lām) making all the difference (Tarū Jaibāl = Trilocanapāla = As everybody knows, the dots in Persian writing often go astray (see Elliot, , ii, 463Google Scholar). Elliot's commentary, for want of the key supplied by the inscriptions, is a bewildering mass of confusion. The statement of that Trilocanapāla had often fled before the Sultan is sufficiently accurate when we remember that the son must have shared his father's flight in the year preceding. The transfer of the Parihār capital from Kanauj to Bārī is deposed to by Alberūnī (Elliot, , i, 54;Google Scholar ii, 464; and Sachau's transl.).

^{page 281 note 1} Cunningham, (Coins Med. India, p. 60Google Scholar) confounds Trilocanapāla of Kanauj (1019–? 1030 A.D.) with the king of the same name belonging to the Ṣāhiya dynasty of Ohind (Kābul), who was defeated decisively by Maḥmūd on the bank of the Tosi River about 1013 A.D., and died in 1021–2 A.D. = 412 H. Trilocana (“three-eyed”) is a name or epithet of Śiva.